Background: Ovarian cancer is characterized by high mortality rates, primarily due to diagnosis at late stages. Current biomarkers, such as CA125, have demonstrated limited efficacy for early detection. While high-dimensional proteomics offers a more comprehensive view of systemic biology, the analysis of such data, where the number of features far exceeds the number of samples, presents a significant computational challenge. Methods: This study utilized a nested case–control cohort of longitudinal pre-diagnostic serum samples from the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) profiled for eight candidate ovarian cancer biomarkers (CA125, HE4, PEBP4, CHI3L1, FSTL1, AGR2, SLPI, DNAH17) and 92 additional cancer-associated proteins from the Olink Oncology II panel. We employed a Synolitic Graph Neural Network framework that transforms high-dimensional multi-protein data into sample-specific, interconnected graphs using a synolitic network approach. These graphs, which encode the relational patterns between proteins, were then used to train Graph Neural Network (GNN) models for classification. Performance of the network approach was evaluated together with conventional machine learning approaches via 5-fold cross-validation on samples collected within one year of diagnosis and a separate holdout set of samples collected one to two years prior to diagnosis. Results: In samples collected within one year of ovarian cancer diagnosis, conventional machine learning models—including XGBoost, random forests, and logistic regression—achieved the highest discriminative performance, with XGBoost reaching an ROC-AUC of 92%. Graph Convolutional Networks (GCNs) achieved moderate performance in this interval (ROC-AUC ~71%), with balanced sensitivity and specificity comparable to mid-performing conventional models. In the 1–2 year early-detection window, conventional model performance declined sharply (XGBoost ROC-AUC 46%), whereas the GCN maintained robust discriminative ability (ROC-AUC ~74%) with relatively balanced sensitivity and specificity. These findings indicate that while conventional approaches excel at detecting late pre-diagnostic signals, GNNs are more stable and effective at capturing subtle early molecular changes. Conclusions: The synolitic GNN framework demonstrates robust performance in early pre-diagnostic detection of ovarian cancer, maintaining accuracy where conventional methods decline. These results highlight the potential of network-informed machine learning to identify subtle proteomic patterns and pathway-level dysregulation prior to clinical diagnosis. This proof-of-concept study supports further development of GNN approaches for early ovarian cancer detection and warrants validation in larger, independent cohorts.

High-dimensional tabular data are common in biomedical and clinical research, yet conventional machine learning methods often struggle in such settings due to data scarcity, feature redundancy, and limited generalization. In this study, we systematically evaluate Synolitic Graph Neural Networks (SGNNs), a framework that transforms high-dimensional samples into sample-specific graphs by training ensembles of low-dimensional pairwise classifiers and analyzing the resulting graph structure with Graph Neural Networks. We benchmark convolution-based (GCN) and attention-based (GATv2) models across 15 UCI datasets under two training regimes: a foundation setting that concatenates all datasets and a dataset-specific setting with macro-averaged evaluation. We further assess cross-dataset transfer, robustness to limited training data, feature redundancy, and computational efficiency, and extend the analysis to a real-world ovarian cancer proteomics dataset. The results show that topology-aware node feature augmentation provides the dominant performance gains across all regimes. In the foundation setting, GATv2 achieves an ROC-AUC of up to 92.22 (GCN: 91.22), substantially outperforming XGBoost (86.05), $\alpha =0.001$. In the dataset-specific regime, GATv2, combined with minimum-connectivity filtering, achieves a macro ROC-AUC of 83.12, compared to 80.28 for XGBoost. Leave-one-dataset-out evaluation confirms cross-domain transfer, with an ROC-AUC of up to 81.99. SGNNs maintain ROC-AUC around 85% with as little as 10% of the training data and consistently outperform XGBoost in more extreme low-data regimes, $\alpha =0.001$. On ovarian cancer proteomics data, foundation training improves both predictive performance and stability. Efficiency analysis shows that graph filtering substantially reduces training time, inference latency, and memory usage without compromising accuracy. Overall, these findings suggest that SGNNs provide a robust and scalable approach for learning from high-dimensional, heterogeneous tabular data, particularly in biomedical settings with limited sample sizes.

In this study, we present a systematic evaluation of Synolitic Graph Neural Networks (SGNNs), a novel framework that transforms high-dimensional tabular data into sample-specific graphs using ensembles of low-dimensional pairwise classifiers. We demonstrate that augmenting these graphs with topology-aware node descriptors (such as degree, strength, closeness, and betweenness centrality) and applying graph sparsification techniques, either via minimum spanning connectivity or fixed-probability edge retention, can significantly improve classification performance. We evaluate both convolution-based (GCN) and attention-based graph neural networks (GATv2) across two training regimes: a foundation model setting where multiple datasets are concatenated, and dataset-specific training. Results show that attention-based models generally exhibit superior performance across classification tasksвАФin the foundation regime, dense (non-sparsified) graphs with node features yield 92.83 ROC-AUC for GATv2 and 92.34 for GCN (vs. 90.80 for XGBoost), and in the dataset-specific regime, GATv2 with minimal connectivity and node features reaches 88.96 ROC-AUC (vs. 86.84 for XGBoost). A leave-one-dataset-out evaluation further indicates out-of-domain transfer to previously unseen datasets (mean ROC-AUC: 0.78 with node features; 0.71 with maximum-threshold sparsification; 0.70 without features). Importantly, we demonstrate that the SGNN framework is capable of overcoming the curse of dimensionality, outperforming traditional machine learning models such as XGBoost in scenarios where the number of features exceeds the number of training samplesвАФmaintaining ROC-AUC above 80\% with only 5\% of the training data, while XGBoost drops to 60\%. Furthermore, SGNNs exhibit robustness to feature redundancy and correlation, with duplicating all features and adding noise producing only minor deviations, reducing the need for manual feature engineering or dimensionality reduction. Across all settings, SGNNs enhanced with node features consistently outperform XGBoost baselines, underscoring the effectiveness of integrating graph-based structural representations, topology-aware augmentation, and controlled sparsification in classification tasks.

This study addresses the critical issue of reliability for AI-assisted medical diagnosis. We focus on the selection prediction approach that allows the diagnosis system to abstain from providing the decision if it is not confident in the diagnosis. Such selective prediction (or abstention) approaches are usually based on the modeling predictive uncertainty of machine learning models involved. This study explores uncertainty quantification in machine learning models for medical text analysis, addressing diverse tasks across multiple datasets. We focus on binary mortality prediction from textual data in MIMIC-III, multi-label medical code prediction using ICD-10 codes from MIMIC-IV, and multi-class classification with a private outpatient visits dataset. Additionally, we analyze mental health datasets targeting depression and anxiety detection, utilizing various text-based sources, such as essays, social media posts, and clinical descriptions. In addition to comparing uncertainty methods, we introduce HUQ-2, a new state-of-the-art method for enhancing reliability in selective prediction tasks. Our results provide a detailed comparison of uncertainty quantification methods. They demonstrate the effectiveness of HUQ-2 in capturing and evaluating uncertainty, paving the way for more reliable and interpretable applications in medical text analysis.

Cardiovascular diseases remain the leading cause of mortality, and automated electrocardiogram (ECG) analysis can ease clinical workloads but is limited by scarce and imbalanced data. Synthetic ECG can mitigate these issues, and while most methods use Generative Adversarial Networks (GANs), recent work show variational autoencoders (VAEs) perform comparably. We introduce cNVAE-ECG, a conditional Nouveau VAE (NVAE) that generates high-resolution, 12-lead, 10-second ECGs with multiple pathologies. Leveraging a compact channel-generation scheme and class embeddings for multi-label conditioning, cNVAE-ECG improves downstream binary and multi-label classification, achieving up to a 2% AUROC gain in transfer learning over GAN-based models.